UTF8PROC_DLLEXPORT utf8proc_ssize_t utf8proc_normalize_utf32()

in subversion/libsvn_subr/utf8proc/utf8proc.c [558:668]

• 108 lines of code
• 44 McCabe index (conditional complexity)

UTF8PROC_DLLEXPORT utf8proc_ssize_t utf8proc_normalize_utf32(utf8proc_int32_t *buffer, utf8proc_ssize_t length, utf8proc_option_t options) {
  /* UTF8PROC_NULLTERM option will be ignored, 'length' is never ignored */
  if (options & (UTF8PROC_NLF2LS | UTF8PROC_NLF2PS | UTF8PROC_STRIPCC)) {
    utf8proc_ssize_t rpos;
    utf8proc_ssize_t wpos = 0;
    utf8proc_int32_t uc;
    for (rpos = 0; rpos < length; rpos++) {
      uc = buffer[rpos];
      if (uc == 0x000D && rpos < length-1 && buffer[rpos+1] == 0x000A) rpos++;
      if (uc == 0x000A || uc == 0x000D || uc == 0x0085 ||
          ((options & UTF8PROC_STRIPCC) && (uc == 0x000B || uc == 0x000C))) {
        if (options & UTF8PROC_NLF2LS) {
          if (options & UTF8PROC_NLF2PS) {
            buffer[wpos++] = 0x000A;
          } else {
            buffer[wpos++] = 0x2028;
          }
        } else {
          if (options & UTF8PROC_NLF2PS) {
            buffer[wpos++] = 0x2029;
          } else {
            buffer[wpos++] = 0x0020;
          }
        }
      } else if ((options & UTF8PROC_STRIPCC) &&
          (uc < 0x0020 || (uc >= 0x007F && uc < 0x00A0))) {
        if (uc == 0x0009) buffer[wpos++] = 0x0020;
      } else {
        buffer[wpos++] = uc;
      }
    }
    length = wpos;
  }
  if (options & UTF8PROC_COMPOSE) {
    utf8proc_int32_t *starter = NULL;
    utf8proc_int32_t current_char;
    const utf8proc_property_t *starter_property = NULL, *current_property;
    utf8proc_propval_t max_combining_class = -1;
    utf8proc_ssize_t rpos;
    utf8proc_ssize_t wpos = 0;
    utf8proc_int32_t composition;
    for (rpos = 0; rpos < length; rpos++) {
      current_char = buffer[rpos];
      current_property = unsafe_get_property(current_char);
      if (starter && current_property->combining_class > max_combining_class) {
        /* combination perhaps possible */
        utf8proc_int32_t hangul_lindex;
        utf8proc_int32_t hangul_sindex;
        hangul_lindex = *starter - UTF8PROC_HANGUL_LBASE;
        if (hangul_lindex >= 0 && hangul_lindex < UTF8PROC_HANGUL_LCOUNT) {
          utf8proc_int32_t hangul_vindex;
          hangul_vindex = current_char - UTF8PROC_HANGUL_VBASE;
          if (hangul_vindex >= 0 && hangul_vindex < UTF8PROC_HANGUL_VCOUNT) {
            *starter = UTF8PROC_HANGUL_SBASE +
              (hangul_lindex * UTF8PROC_HANGUL_VCOUNT + hangul_vindex) *
              UTF8PROC_HANGUL_TCOUNT;
            starter_property = NULL;
            continue;
          }
        }
        hangul_sindex = *starter - UTF8PROC_HANGUL_SBASE;
        if (hangul_sindex >= 0 && hangul_sindex < UTF8PROC_HANGUL_SCOUNT &&
            (hangul_sindex % UTF8PROC_HANGUL_TCOUNT) == 0) {
          utf8proc_int32_t hangul_tindex;
          hangul_tindex = current_char - UTF8PROC_HANGUL_TBASE;
          if (hangul_tindex >= 0 && hangul_tindex < UTF8PROC_HANGUL_TCOUNT) {
            *starter += hangul_tindex;
            starter_property = NULL;
            continue;
          }
        }
        if (!starter_property) {
          starter_property = unsafe_get_property(*starter);
        }
        if (starter_property->comb_index < 0x8000 &&
            current_property->comb_index != UINT16_MAX &&
            current_property->comb_index >= 0x8000) {
          int sidx = starter_property->comb_index;
          int idx = (current_property->comb_index & 0x3FFF) - utf8proc_combinations[sidx];
          if (idx >= 0 && idx <= utf8proc_combinations[sidx + 1] ) {
            idx += sidx + 2;
            if (current_property->comb_index & 0x4000) {
              composition = (utf8proc_combinations[idx] << 16) | utf8proc_combinations[idx+1];
            } else
              composition = utf8proc_combinations[idx];

            if (composition > 0 && (!(options & UTF8PROC_STABLE) ||
                !(unsafe_get_property(composition)->comp_exclusion))) {
              *starter = composition;
              starter_property = NULL;
              continue;
            }
          }
        }
      }
      buffer[wpos] = current_char;
      if (current_property->combining_class) {
        if (current_property->combining_class > max_combining_class) {
          max_combining_class = current_property->combining_class;
        }
      } else {
        starter = buffer + wpos;
        starter_property = NULL;
        max_combining_class = -1;
      }
      wpos++;
    }
    length = wpos;
  }
  return length;
}